210 The Basics of financial economeTrics
Panel B of Table 10.7 presents the error-correction model estimates
for each of the three countries. The error-correction term in each equation
reflects the deviation from the long-run stochastic trend of that stock index
in the last period. It should be noted that in contrast to the Engle-Granger
cointegration test, the Johansen-Juselius cointegration test estimates the
long-run and short-run dynamics in one step. The speed of adjustment
coefficient provides insight into the short-run dynamics. This coefficient is
insignificant (at the 5% level) for Germany. This means that stock prices in
Germany do not change in response to deviations from their stochastic trend
with France and the Netherlands. Because the variables are cointegrated,
we are guaranteed that at least one speed of adjustment coefficient will
be significant. In fact, the speed of adjustment coefficients of both France
table 10.7 Cointegration Equation and Error Correction Equations, 1999− 2007
Panel A. Cointegrating Equation
France = 4.82 + 2.13 Germany − 1.71 Netherlands
[8.41] [5.25]
Panel B. Error-Correction Equations
Country A(France) A(Germany) A(Netherlands)
Zt− 1 −0.151477
[−2.21]
−0.057454
[−0.66]−0.179129
[−2.52]Δ(France(−1)) 0.087360
[0.27]
0.245750
[0.60]0.225357
[0.67]Δ(France(−2)) −0.200773
[−0.68]
−0.218331
[−0.58]−0.324250
[−1.06]Δ(Germany(−1)) −0.189419
[−0.82]
−0.024306
[−0.08]−0.094891
[−0.39]Δ(Germany(−2)) −0.155386
[−0.67]
−0.109070
[−0.37]−0.127301
[−0.53]Δ(Netherlands(−1)) 0.079881
[0.34]
−0.189775
[−0.64]−0.188295
[−0.77]Δ(Netherlands(−2)) 0.439569
[1.89]
0.446368
[1.52]0.483929
[2.00]C 0.005967
[1.02]
0.002575
[0.35]0.002688
[0.44]France(−1) represents the log return of the French stock index from the previous
month. Germany(−1) and Netherlands(−1) have a similar interpretation. Numbers
in brackets represent the t-statistic.